TLDR AI · ENGINEERING & RESOURCES

SCIENCE AGENT BENCH

github.comGitHub repository

Summary from the TLDR AI newsletter

Benchmark code for the science agent test. This essentially serves to measure how good models are at producing novel scientific advances.